ALPR Audit Watch

Under Minn. Stat. § 13.824, subdivision 6, any law enforcement agency operating an automated license plate reader must arrange an independent, biennial audit of its ALPR records. The audit has to confirm that the data is properly classified, that it is being used only as the statute allows, that it is destroyed on schedule, and that every query was authorized.

Results go to the Commissioner of Administration, the chairs and ranking minority members of the relevant legislative committees, and the Legislative Commission on Data Practices, within 30 days. If an audit finds substantial noncompliance, the commissioner may order the agency’s readers shut off until it complies.

Separately, ALPR data unrelated to an active investigation must be destroyed within 60 days, and agencies must report each reader to the BCA within ten days of installation.

The statute does not define who counts as an independent auditor, what the audit must test, or what the report must contain. In practice a handful of firms and consortiums produce most of the reports in Minnesota, several of them run by former law enforcement officials, and several of them auditing the same member agencies that fund them.

A real audit almost always turns something up. Records get held past the 60-day deletion window, a reader goes unreported to the BCA, a log of use never gets maintained, a query lacks documented authorization. Findings are the visible evidence that someone actually looked. An auditor who reports them is doing the job the legislature described.

A template that clears every agency it touches is a different thing. When the same firm files the same checklist year after year and never records a discrepancy, the audit stops being an inspection and becomes a stamp. That satisfies the letter of the statute while defeating its purpose, which was to put an independent set of eyes on a surveillance system the public cannot see.

Of the 68 audits we have read in full, 56 recorded zero findings (82% of them). The reports run from two pages to nine. Most are checklists: a statute-by-statute table, a policy review, and a sentence stating that no discrepancies were noted.

The audits that do dig in find real problems. The one auditor who tested the live platform found that Flock data exports generate no audit-trail entry at all, a flaw that undercuts the 60-day deletion rule the audit exists to verify, and one that applies to every agency running the same system. Nobody else caught it.